Towards the Unification and Robustness of
              Perturbation and Gradient Based Explanations

           Sushant Agarwal 1 Shahin Jabbari 2 Chirag Agarwal 2 * Sohini Upadhyay 2 *
                  Zhiwei Steven Wu 3 Himabindu Lakkaraju 2

              Abstract                  can understand and consequently trust their functionality.
   As machine learning black boxes are increasingly         Only if decision makers have a clear picture of the behav-
   being deployed in critical domai ...